QMIX: Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning
The StarCraft Multi-Agent Challenge : Environment Code
The StarCraft Multi-Agent Challenge : Research Paper
Using Pytorch 1.3.
Anaconda.
Windows 10.
Be sure to set up the environment variable : SC2PATH (see lauch.bat)
python train.py --scenario [scenario_name] --train
or
launch.bat
python train.py --scenario [scenario_name] --load-episode-saved [episode number]
or
launch eval.bat
This will generate a SC2Replay file in {SC2_PATH}/Replays/replay